Does a States Use of Indiscriminate Violence Incite Insurgent Attacks - Article Example

LYALL’S ARTICLE Introduction Does a s use of indiscriminate violence incite insurgent attacks? To most existing theories and empirical studies have concluded that such violence is highly counterproductive because it creates new grievances while forcing victims to seek security, if not safety, in rebel arms. This proposition is tested using Russian artillery fire in Chechnya (2000 to 2005) to estimate indiscriminate violence’s effect on subsequent patterns of insurgent attacks across matched pairs of similar shelled and nonshelled villages. The findings are counterintuitive. Shelled villages experience a 24 percent reduction in posttreatment mean insurgent attacks relative to control villages. In addition, commonly cited “triggers” for insurgent retaliation, including the lethality and destructiveness of indiscriminate violence, are either negatively correlated with insurgent attacks or statistically insignificant. Main argument & Research question This article uses Russian artillery strikes on populated settlements in Chechnya (2000 to 2005) to test the presumed relationship between a state’s indiscriminate violence and insurgent attacks. This shelling offers an identification strategy that allows the researcher to compare levels of insurgent violence before and after an artillery strike in a shelled village with those of a similar but not victimized village during identical time frames. Contrary to most existing studies, this difference-in-difference estimation finds that indiscriminate violence actually reduced the mean number of insurgent attacks relative to nonshelled villages. Moreover, commonly cited “triggers” for insurgent attacks, including the number of casualties inflicted and the amount of property damage suffered, are either negatively correlated with insurgent violence or statistically insignificant. As an insurgent organization’s ranks swell, so too does its coercive capacity. More specifically, it is reasonable to assume that insurgent organizations with larger memberships are more capable of planning and conducting a greater number of attacks than their smaller counterparts. This is likely to be true independent of the skill level of the particular insurgents in question. For example, while organizations with fewer combatants may be able to match the output of larger organizations temporarily, their long-term ability to do so is questionable since sustaining such a rate of violence means demanding more attacks per insurgent, thus exposing these insurgents to relatively higher risk of death or capture. Indiscriminate violence thus appears doubly counterproductive: it not only helps alleviate the insurgent’s collective action dilemma but also increases the amount of “action” that an organization can generate and sustain over time. Hypotheses Insurgent leaders can therefore find themselves in a severe bind: if they continue to mount attacks from within (or near) an aggrieved population, they risk (further) alienation of disillusioned locals, raising the specter of defection to the incumbent’s side; if they curb their violence in recognition of popular pressures, they risk introducing inefficiencies into their strategy by reducing their war-fighting capabilities. Indiscriminate violence thus reshapes the relationship between insurgents and populace by underscoring that the insurgency cannot credibly protect the population, and moreover, that its continued presence endangers noncombatants. Without an adequate response to state violence, insurgents are likely to be perceived as the weaker side, thus removing an important incentive for joining the insurgency. Research design: variable, data and procedures I use artillery strikes by Russian forces on populated settlements in Chechnya (2000 to 2005) as an identification strategy to isolate the causal effect of indiscriminate violence. These strikes have an important property: they are uncorrelated with key spatial and demographic variables thought to drive insurgent-attack propensity, including population size, presence of incumbent bases, and terrain. Yet, the most important determinants of insurgent violence may be war-induced dynamics that arise out of the interaction of Russian and rebel strategy. To account for this possibility, I adopt matching to create populations of shelled (“treated”) and nonshelled villages that are similar across background covariates as well as prior Russian and insurgent military activities. Difference-in-difference estimation is then used to gauge shelling’s effect by comparing mean differences in insurgent attacks before and after a strike across shelled and control villages during identical time periods. The second Chechen War represents a “most likely” case (Eckstein 1975) for observing the link between indiscriminate repression and increased insurgent attacks. Indeed, the war has witnessed astonishing levels of brutality by all sides and has often been described in escalatory terms as each side’s violence radicalized the other’s tactics and aims (Hahn 2007; Wilhelmsen 2005). “Chechnya,” one observer has concluded, “is above all a lesson in the devastating spiraling dynamic of violence” (Zürcher 2007, 113). The conflict has become synonymous with excesses by Russian forces and their pro-Russian Chechen allies. Human Rights Watch, the European Court of Human Rights, and local nongovernmental organizations (NGOs) have issued a stream of reports decrying the use of indiscriminate violence by these forces, including artillery and air strikes on populated places. Village-sweep operations (zachistki) are routinely marked by forced disappearances (about five thousand since 1999) and extrajudicial killings (Human Rights Watch 2002a, 2002b, 2006). Insurgents have retaliated with suicide bombings, mass hostage-takings, and a relentless campaign of attacks against Russian patrols. An estimated fifteen thousand to twenty-five thousand civilians and at least five thousand Russian soldiers have died since 1999; roughly one hundred thousand citizens were also temporarily internally displaced. As a “small corner of Hell” (Politkovskaya 2003), Chechnya would appear a clear example of indiscriminate repression fueling an insurgency. When attempting to isolate the causal impact of indiscriminate violence, we need to take stock of as many potentially confounding variables as possible. This section details eight demographic, spatial, and conflict-related variables commonly used to explain patterns of insurgent violence. All measures were taken before the artillery shelling (June 2000 to December 2005) that acts as our treatment. Population records the number of inhabitants in a populated settlement (logged) in March–April 2000, when the first comprehensive household survey was conducted (Danish Refugee Council 2000). Poverty uses a threefold classification scheme to rank the severity of a population’s need for humanitarian assistance. Elevation records a village’s altitude in meters (logged). Isolation measures the number of settlements that are found within 5 km2 of the swept village. This captures the belief held among practitioners that isolated villages are easier to suppress because insurgents have few or no options when seeking to escape or hide within the local populations (U.S. Army Field Manual 2007, 185). Garrison demarcates whether a Russian garrison was stationed in a particular village by June 1, 2000. Finally, rebel records whether a village was located in a district controlled by or aligned with Shamil Basayev or Doku Umarov. These leading rebel commanders were guided by different ideologies, and as a result, pitched their recruitment appeals around either radical Islamic tenets (Basayev) or nationalism (Umarov). An average decrease of .51 attacks may appear to be a substantively small effect. The cumulative effect is quite large, however. With at least 336 attacks identified in the cumulative preshelling windows of treated villages, about 81 attacks are missing from the posttreatment interval because of artillery strikes (or between 28 and 136 missing attacks with a 95 percent confidence interval). More bluntly, the average insurgent attack in the cumulative pretreatment window of shelled villages killed 0.88 Russian soldiers and pro-Russian Chechen militia members and wounded another This reduction in insurgent attacks therefore translates into about seventyone soldiers who avoided being killed by insurgent violence (or between 25 and 120 with a 95 percent confidence interval), along with a further 107 soldiers (or 34 to 165, with a 95 percent confidence interval) who escaped being wounded. These figures lay bare the brutal logic behind indiscriminate violence: it can suppress insurgent violence, and in so doing, degrade insurgents’ military capabilities. Effects or non effect and treatment Existing theories also implicitly suggest that state violence works through several mechanisms (or “triggers”) to change patterns of insurgent violence. It is often assumed, for example, that a positive relationship exists between the lethality and destructiveness of state actions and the amount of violence subsequently generated by insurgents. This section explicitly tests this relationship by moving beyond the binary treated–control distinction to examine various facets of the treatment itself. More specifically, three aspects of the treatment are examined here: its lethality (the number of individuals killed and wounded in a strike); its destructiveness (the number of buildings and farms damaged or destroyed in a strike); and the frequency of shelling (the number of times a village was struck). I therefore distinguished between artillery strikes that killed individuals, wounded individuals, and destroyed property from those that did not using dummy variables. I also created a dummy variable to reflect whether a village had been shelled more than once. I also created variables with the actual count for each type of violence inflicted on settlement inhabitants; for property damage and multiple strikes, I logged these values to reduce skew and kurtosis. Several findings emerge. First, all but one of these eight measures of the type and severity of violence inflicted are negatively associated with increases in postshelling insurgent violence. Setting aside the dummy variable for property damage, every measure contradicts the contention that state actions provide a catalyst for increased insurgent violence. Second, only two measures—the dummy variables for deaths inflicted and for wounded individuals—reach conventional levels of significance. These findings suggest that insurgent retaliation may vary in part on the type of violence inflicted by the state, with lethality clearly triggering a (negative) response not mirrored by either property damage or frequency of shelling. Interestingly, little evidence appears to exist for the contention that higher levels of violence inflicted (as captured in the count variables) are strongly associated with either increased or decreased insurgent violence. We must be cautious in interpreting this finding, however, since the small number of observations available here can provide only an initial test of the relationship between severity of indiscriminate violence and observed insurgent responses. Testing these relationships in multiple contexts, as well as sorting out which mechanisms are at work (and why), will be a close-range task by necessity: nationallevel data are simply too crude to capture these dynamics. Even case-based methods may prove inadequate if the often-severe inferential threats that lurk in civil war data are not addressed with appropriate research designs. These efforts will demand both conflict-specific knowledge and methodological skills but hold out the promise of substantially enriching our theories of the dynamics of violence in civil wars. Questions The questions that this article raises for us in the West however are very important. How much should we take this message to heart in the war in Afghanistan, for example, or in future conflicts? At what price should victory be sought? Should we be more brutal when we fight wars? How well does this strategy compare to a “win their hearts and minds” strategy? Is the fact that we appear to have lost momentum in Afghanistan a consequence of our refusal to fully engage the enemy? These are difficult questions that are raised implicitly by Lyall’s article and must be addressed by policymakers.   Read More